AN EMPIRICAL STUDY ON PRICE DISCOVERYstrong, semi-strong, and weak form efficiency (Fama, 1970)....
Transcript of AN EMPIRICAL STUDY ON PRICE DISCOVERYstrong, semi-strong, and weak form efficiency (Fama, 1970)....
*Undergraduate student, Amherst College, Amherst, MA 01002. Tel: (504)905-7341. E-mail: [email protected]
AN EMPIRICAL STUDY ON PRICE DISCOVERY
IN THE HONG KONG EQUITY MARKET
SHI YUAN CHEN *
SUBMITTED TO THE DEPARTMENT OF ECONOMICS OF AMHERST COLLEGE
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
BACHELOR OF ARTS WITH HONORS
PROFESSOR STEVEN RIVKIN, FACULTY ADVISOR
MAY 8, 2008
2
ACKNOWLEDGEMENTS
This undergraduate Economics thesis is my first attempt at utilizing four years of studies primarily in Economics, Mathematics, and Computer Science at Amherst College. I would first like to thank my thesis advisor, Professors Steven Rivkin, without whose guidance and encouragement this work could never have been completed. In addition, Professor Geoffrey Woglom provided critical help in the early stages of developing my thesis idea. I would like to thank my parents and my sister for their tremendous support during my time at Amherst. I would like to sincerely thank Professor Jun Ishii for discovering a fatal flaw with my time series dataset only days before the due date. Although the fixing process involved a great deal of scrambling, I am happy with the way the paper turned out relative to what it could have been otherwise. Thanks to everyone else for making this thesis journey exciting and bearable.
3
ABSTRACT
This thesis investigates the impact of two separate events on the Hong Kong Stock
Exchange (HKEX): the opening of the stock options market in Hong Kong in 1995 and
the announcement that the exchange would be open to mainland Chinese investors in
2007. In the first case, the Hang Seng Index is tested for evidence refuting random walk
hypothesis using the Geary run test. This test was run on 8-year periods before and after
the opening of the options market. The run test detected evidence of serial dependence
which can be interpreted as the market being predictable or inefficient. In the second
case, I examine the price relationship between shares of Chinese companies traded on
both the Hong Kong stock exchange and the New York stock exchange. Using recent
daily data from 2003 through 2008 across fourteen companies, I first checked to see if the
stock price time series were stationary by running the Dickey Fuller test. The stock
prices were found to be nonstationary, but the first differences of the stock prices or
return series were stationary. This evidence of cointegration allowed a Granger causality
test to be run on the return prices of two stocks. I found that price discovery existed
between HKEX and NYSE as one would expect. I subsequently tested for whether or not
there were any persistent price disparities between US ADRs and Hong Kong H- shares
shortly after the announcement was made. I found that although all of the stocks
experienced a price increase of 10% to 30% during the week of the announcement, there
were no arbitrage opportunities between the US and Hong Kong markets.
Keywords: Information and Market Efficiency; Hong Kong Stock Exchange; Adaptive Markets Hypothesis
4
TABLE OF CONTENTS
Acknowledgements………………………………………………………………… ii
Abstract…………………………………………………………………………….. iii
1. INTRODUCTION……........................................................................................ 5
2. THEORETICAL FOUNDATIONS………………..……………………………… 7
2.1. Motivation…………………………………………………………….. 8
2.2. Efficient Markets Hypothesis………………………………………… 9
2.3. Model Anomalies……………………………………………………... 11
2.4. Adaptive Markets Hypothesis………………………………………… 13
2.5. Stock Price Discovery……………………………………………....... 15
3. DATA………………………………………………………………………. 18
4. METHODOLOGY…………………………………………………………….. 22
4.1. Geary Run Test………………………………………………………. 22
4.2. Granger Causality Test……………………………………………..… 23
4.3. Arbitrage Trading Rule……………………………………………….. 25
5. EMPIRICAL RESULTS AND ANALYSIS……………………………………...... 25
5.1. Stock Options Market…...………………………………………..…… 25
5.2. Announcement Effect…………………………………………………. 26
6. CONCLUSION………………………………………………………………...30
References…………………………………………………………………………. 32
Appendix…………………………………………………………………………… 36
5
I. INTRODUCTION
The famous American economist James Tobin once stated that the most important
implication of the efficient markets hypothesis is the channeling of investment to the
most efficient users of capital (Tobin, 1982). Under this generally accepted framework,
inefficiency in the stock market results in poor business investment decisions, imparting a
real cost onto the economy. In the past thirty years, the consensus among economists has
been that stock prices follow a random walk process and that the stock market is semi-
strongly efficient under the efficient markets hypothesis proposed by Eugene Fama
(Fama, 1970). Recently however, a study has found evidence of weekly returns of the
US stock market rejecting the random walk hypothesis (Lo and MacKinlay, 1988). The
focus of this paper is to survey the price behavior and efficiency of the Hong Kong stock
exchange in response to important events.
The Hong Kong stock market has developed rapidly since its establishment in the
late 1800s. It has always played a large role in facilitating trade between China and the
rest of the world, and this continues to be true in the capital markets today. It is currently
the third largest stock market in Asia and the sixth largest in the world. For more than a
decade, Hong Kong has been recognized for closely following free-market principles; the
Wall Street Journal gave Hong Kong the title of World’s Freest Economy for the 13th
consecutive year based on a ranking system that considers several factors such as trade
freedom, investment freedom, financial freedom, and property rights.
The two events this paper looks at are the launch of the Hong Kong stock options
market on September 8th, 1995 and China’s announcement on August 20th, 2007 that the
Hong Kong stock market would finally become accessible to the millions of investors in
6
mainland China. The opening of the options market in the Hong Kong stock exchange
coincided with a regulatory decision to allow naked short selling in 1995. I expect that
the level of efficiency in the stock market will rise in the periods following 1995 due to
these additional investment options and changes in regulation. China’s announcement in
2007 gave investors a place to invest their money outside of the Shanghai stock exchange
and the Shenzhen stock exchange, both of which have inefficient prices due to constraints
on short-sales. A study found that short-sales constraints in stock markets tend to cause
stock price overvaluation (Chang, 2004). Because of the high premium on shares traded
in Shanghai compared to Hong Kong and the US, I surmise that the opening of the Hong
Kong market to Chinese investors created inefficiencies and possibly arbitrage
opportunities for global investors. An estimated $100 billion USD of capital inflow
within the first year alone would account for 3-4% of expected growth to the market
value of the Hong Kong Stock Exchange.
Although mainland Chinese investors did not have direct access to the Hong
Kong market until late 2007, anticipation of the entry of optimistic investors would have
had an effect on the stock prices at announcement. Given a relatively efficient pricing
structure in the free-market Hong Kong stock exchange and an overvalued prices in the
restricted Shanghai stock exchange, it is not entirely obvious what would happen to
prices when investors are allowed to move from inefficient markets to efficient ones.
In the second half of the paper, I discuss my testing for the following hypotheses:
1) there are long-term cointegrations between the prices of Hong Kong-traded shares and
US-traded shares of the same company and 2) the opening of the market to Chinese
investors in 2007 caused a period of high price volatility. As a result, the market faced
7
temporary inefficiencies and possible arbitrage opportunities that disappeared once the
market caught on. In order to test for cointegration, the Granger causality test requires
that the set of underlying time series be stationary. The Dickey-Fuller test, the standard
test for autocorrelation, was used to detect whether or not the time series dataset had unit
root (Dickey and Fuller, 1979). I then ran the Granger causality test on the first
difference series or return time series which was found to be stationary. Finally, I
simulated an arbitrage trading rule in an effort to detect if profitable opportunities
materialized immediately after the announcement.
Section 2 discusses the theoretical foundations underlying the application and
evaluation of the tests for price relationship. The section begins with a discussion of the
efficient markets hypothesis and adaptive markets hypothesis followed by a summary of
the characteristics of the daily price data collected from the Hong Kong and New York
stock markets. Sections 3, 4 and 5 discuss the data set, methodology, and empirical
results respectively. Section 6 concludes the paper with a summary of my results from
testing the dual market and its importance in understanding price discovery in the Hong
Kong stock exchange.
II. THEORETICAL FOUNDATIONS
Despite its popularity in academic circles and its heavy influence on modern investment
theory, the Efficient Markets Hypothesis (EMH) is not a view deeply embraced by many
investors actually participating in the financial markets. The origins of the efficient
markets hypothesis dates back to 1965 when Samuelson published his proof that properly
8
anticipated prices fluctuate randomly (Samuelson, 1965). The term “efficient market”
was first introduced into economics literature by Fama et al. with three classifications:
strong, semi-strong, and weak form efficiency (Fama, 1970).
Under the efficient markets hypothesis, any kind of investment strategy based
solely on historical data will not consistently beat the market except through pure luck.
Most experienced investors would disagree with this point, due to their empirical
observations of market pricing biases or anomalies. In fact, investors would not waste
time performing research on investments if they believed that the market was efficient;
they must presume that markets are not efficient. Grossman and Stiglitz argue that
efficient markets with perfect information are an impossibility, because if there was no
profit in gathering information, there would no longer be any incentive to gather
information thus leading to inefficiencies (Grossman, 1980). They proposed that the
degree of market inefficiency determines investors’ efforts to gather and trade on
information.
This section begins with an explanation as to how inefficiency in the stock market
would affect the real world in terms of the allocation of capital. This is followed by a
presentation of the efficient markets hypothesis framework and a summary of the
empirical findings of anomalies in market pricing. I then turn to the theory of adaptive
markets and its applicability to markets. Finally, I discuss the theory behind price
discovery of Chinese ADRs.
2.1. Motivation. The stock market is important to the study of economics because it
guides corporate investment by transferring two kinds of information: information about
9
investment opportunities and information about managers’ past decisions. Capital
budgeting theory states that in evaluating investment projects, expected cash flows are
discounted at risk-adjusted rates of return. Rather than undertake the costly effort to
determine the value of the project, the manager can rely on the company’s stock price
movement to judge whether or not it has positive Net Present Value. The underlying
reasoning is that traders will seek to realize this profit opportunity by attaining and
analyzing the necessary data. Specifically, information about similar investment projects
in the past is studied by traders in the stock market to determine the discounted risk-
adjusted rates of return. In equilibrium, information in stock prices will guide investment
decisions because managers will be compensated based on informative stock prices in the
future. Under the efficient market hypothesis, the price will move towards this
equilibrium, which reflects the company’s value as if it had already taken on the project.
One study argues that the stock market may not be a necessary institution for the
efficient allocation of equity, because stock prices play only an indirect role in which
investment projects the company decides to take on (Dow and Gorton, 1997).
Nonetheless, it is still a very important signal for managers to use when evaluating an
investment opportunity. Thus, studying the efficiency of the stock market will provide
insight to the overall efficiency of the economy.
2.2. Efficient Markets Hypothesis. As introduced by Fama et al. (Fama, 1970), there
are three types of efficient markets: (1) strong form, (2) semi-strong form, and (3) weak
efficiency.
10
Strong efficiency – This form asserts that security prices reflect all information, both
public and private. In other words, insider information could not give an investor an
advantage.
Semi-strong efficiency – This form of EMH claims that all publicly available
information is reflected in prices. Because both fundamental and technical analyses are
based on publicly available information, neither would be expected to return risk-adjusted
above-market returns.
Weak efficiency – This form implies that only past data is fully reflected in security
prices. Under this form, technical analysis is not expected to beat the market.
The semi-strong form of EMH is generally the basis for most empirical research. When
examining the financial markets, most economists have ruled out the strong form because
of its strong assumptions. Seyhun has shown that insiders are able to beat the market by
trading on private information (Seyhun, 1986; Seyhun, 1998).
Investors and economists seem to hold opposing beliefs regarding the existence of
investment opportunities. Economists believe that, unless barriers restrict the market
from being efficient, investors will compete away inefficiencies until the equilibrium
market price is efficient. Grossman and Stiglitz have argued inefficiencies must exist
because investors would otherwise spend their time on a more productive role. If there
were very little or no profit in gathering information and seeking out inefficiencies, the
market would fall into an inefficient state. Therefore, the level of inefficiency in a single
market must be at the threshold level between what is considered inefficient and efficient.
This means that the risk-adjusted expected return level of the stock market as a whole
11
should be determined by the point where investors are indifferent between investing and
their next best profitable endeavor. On the other hand, investors believe that investment
opportunities that beat the average expected return level do exist and they actively seek to
find them.
This conundrum can be resolved by closely examining the underlying meaning of
efficiency. When economists state that “markets are efficient,” they mean that any given
single market will become efficient in the long run. When investors say that markets are
inefficient, their beliefs are that given the large universe of potential trading strategies in
a continually volatile market, that there will always be an opportunity other investors
have overlooked to make short-term profits. So in fact it is possible for both statements
to be true at the same time. Note also that there is a disparity between the time scales of
economists and investors. Even the term “long-term investor” refers to an investor who
aims to capitalize on opportunities that are short-term relative to the amount of time it
could take for a market to become efficient. The “long-term” economists refer to is more
theoretical and can last a very long time. To avoid potential confusion, any reference in
this paper to “long-term” will always take on the economist’s meaning.
2.3. Model Anomalies. One of the most controversial results of EMH is that efficient
markets do not allow investors to earn above-average expected returns without accepting
above-average risks. Historical evidence supporting the contrary is abundant, and these
examples as classified as asset pricing “anomalies.” Economists would argue that many
of the anomaly findings are a result of data mining. Given a large enough amount of
historical data, there is bound to be some subset that will generate above average risk-
adjusted returns. In addition, economists assert that while these inefficiencies existed,
12
investors would not have known about these opportunities until they disappeared and the
market prices had been corrected. In this section, I will outline some of the most famous
and compelling cases of anomalies found in the US stock market.
A. The January Effect: Studying NYSE stocks between 1904 and 1974, Rozeff and
Kinney (1976) found a higher mean return in January as compared to other months. The
average return in January was 3.48 percent as opposed to only 0.42 percent for all other
months. The same effect was found in more recent data by Bhardwaj and Brooks (1992)
and Eleswarapu and Reinganum (1993) and in other countries by Gultekin and Gultekin
(1983). The authors offer a tax-loss selling explanation for this anomaly.
B. The Monday Effect: French (1980) found a tendency for the returns of U.S. stocks to
be negative on Mondays and positive on other days of the week using data between 1953-
1977. This anomaly strongly supports a more evolutionary view of stock markets since
Kamara (1997) later found that the S&P 500 no longer showed significant Monday
effects after April 1982.
C. The Small Firm Effect: Banz (1981) and Reinganum (1981) analyzed stocks of low
capitalization companies in the period 1936-1975. Reinganum found that risk adjusted
annual return of small firms was greater than 20%.
D. The P/E Ratio Effect: Basu (1977) looked at companies with low price-to-earnings
ratios and found that between 1957 and 1971, these stocks performed significantly better
than market returns. Campbell and Shiller (1988) showed that P/E ratios have a reliable
forecast power.
13
E. The S&P Index Effect: Harris and Gurel (1986) and Shleifer (1986) found that in the
past, news of a stock’s inclusion into the S&P 500 index would significantly increase the
share price. This information is not about the firm itself, so it is surprising to see this
news affect its stock price.
F. The Weather Effect: Saunders (1993) found the NYSE index tended to be negative
when it was cloudy. Hirshleifer and Shumway (2001) analyzed data from 26 countries
from 1982-1997 and found that in almost all of the countries studied, stock market
returns were positively correlated with sunshine and that snow and rain had no predictive
power.
2.4. Adaptive Markets Hypothesis. Most investors and economists agree that there are
no long-term return anomalies in the financial markets (Lakonishok, Shleifer and Vishny,
1994). Their viewpoints differ on whether or not there exist any short-term opportunities
that can be exploited. Recent behavioralist studies such as Shleifer and Vishny (Shleifer
and Vishny, 1990), Odean (1998), Daniel et al (2000), and Hong et al (1999) have found
that there are certain limits to arbitrage that prevent markets from reaching higher levels
of efficiency. As a result of these constraints on arbitrage, individual irrationality can
spread to the market. The psychological causes for irrationality have been well
documented, for example by Barberis and Thaler (2002). Psychologists have found that
people act irrationally due to overconfidence, optimism and wishful thinking,
representativeness, conservatism, belief perseverance, anchoring, and availability biases.
Critics of behavioral finance state that the asset pricing “anomaly” models can only fit a
specific case, but they do not offer a satisfactory general framework of their own. This is
where the adaptive markets hypothesis (Lo, 2004) comes in.
14
Rather than taking the neoclassical approach of attempting to maximize expected
utility and assuming rational expectations, AMH has an evolutionary perspective and
views agents (investors) simply as dynamic organisms that have adapted through
generations of natural selection and have a goal of maximizing returns. Individuals act in
their own self interest, and can make mistakes, learn, and adapt. In addition, competition
drives adaptation and innovation, natural selection shapes market ecology, and evolution
determines market dynamics. Under this fundamental framework, the AMH offers a
viable alternative to EMH while taking into account the biases found in behavioral
finance.
Under AMH, there is a time period between when the first investor finds a
profitable arbitrage opportunity and when the market corrects the inefficiency. Investors
seek out profitable strategies because they believe that there are opportunities which other
investors have not discovered. In a typical noisy stock market where one investor is
aware of the opportunity and no one else is, the discovery of the opportunity will not be
reflected in stock price. The profit an informed investor makes is a function of the
second-highest bid in the market; this is in some ways similar to a Vickrey auction. Once
another investor discovers the successful strategy then the profit opportunity is driven
down to zero assuming both investors have enough capital to fully employ the arbitrage
opportunity.
Neely, Weller, and Ulrich (2007) tested the validity of claims to excess returns by
using technical trading rules in the foreign exchange market (Brock, Lakonishok and
LeBaron, 1992; Sullivan, Timmerman and White, 1994). Using out-of-sample
performance data, they found that excess returns during the period 1970-1990 were
15
genuine and not a result of data mining. By the early 1990s, these opportunities
disappeared, most likely due to a general awareness of these successful strategies among
investors and the adoption of filter and moving average rules. This finding is consistent
with the AMH but not with EMH.
2.5 Stock Price Discovery. In order to closely examine the behavior of two markets for
the same good, it is necessary to discuss the idea of price discovery. Price discovery is
the process by which markets incorporate new information in order to reach equilibrium
prices. Price discovery between two stock markets in which there is no overlapping
period of time where both markets are open can be difficult due to the constant flow of
new information that may affect the value of the stock. Some stock markets that do not
have this problem can still experience no price discovery; recent studies have looked at
price disparities in the Chinese stock markets. Shares traded in on either of the two
mainland Chinese exchanges are termed A- and B- shares and those traded in Hong Kong
are called H-shares. Guo and Tang (2006) examined the price disparity between A- and
H- shares of 29 companies between 1993 and 2003. They attribute the price disparity to
cost of capital and liquidity. Chan and Kwok (2005) examined 13 A/H shares from 1991
through 2000 and attributed the differences in price to liquidity, supply risk, and
information asymmetry. Wang and Li (2003] accredited price disparity of 16 A/H shares
between 1995 and 2001 to liquidity, risk, market conditions, and exchange rate. If we
consider the mainland Chinese to be a developing market and Hong Kong a developed
market, then these findings are in line with another study (Ratner and Leal, 1999) that
found that equity markets in developing countries are significantly less efficient than
those in developed countries.
16
An American Depository Receipt, or ADR, is a financial instrument traded on US
markets that represents ownership in the traded shares of a foreign company. ADRs give
US investors the ability to buy shares of foreign companies without having to purchase
shares overseas. Two banks are generally involved in maintaining an ADR on a US
exchange: an investment bank and a depository bank. The investment bank purchases the
shares and offers them for sale in the US. The depository bank handles the issuance and
cancellation of ADR certificates and sets the ratio of US ADRs per home country share.
It is important to note here that the owner of an ADR has the right to obtain the foreign
stock it represents at the end of the trading day. Because of this right, it is easy in theory
to execute arbitrage if there is a large enough disparity between the two stock prices to
offset arbitrage transaction costs. However, the Hong Kong market is not open when the
US market closes, so arbitrage is actually a little more difficult. This paper discusses
arbitrage opportunities in further detail in the methodology section. I expect to find that
the price of an ADR, adjusted for the exchange rate and the ratio of ADRs to foreign
company shares, should always move closely with the price of the foreign stock.
There are various types of ADR programs that meet the needs of different
companies:
Unsponsored – No regulatory reporting requirements and shares are issued in
accordance with market demand
Restricted Programs – This plan is tailored for companies that wish to limit the access
of their stock only to certain individuals.
17
Level 1 – Shares can only be traded on the OTC market and there are very few reporting
requirements. The program is sponsored, so the foreign company works with an
American bank to act as a transfer agent.
Level 2 – This allows companies to be listed on a U.S. stock exchange such as the NYSE,
NASDAQ, and the American Stock Exchange (AMEX). However, this also requires the
foreign company to follow stricter reporting requirements and SEC regulations.
Level 3 – As the highest level a foreign company can have, this requires adherence to
even stricter US rules and regulations. The listing of this stock involves an offering in
which the company raises capital
This paper examines how price movements in the HKEX and the NYSE differ
after a significant event such the announcement of a large influx of Chinese capital. An
advantage of comparing Hong Kong-listed and US-listed stocks is that the two markets
are so large and similar in restriction that liquidity will not be as large a concern.
I would like to consider what risks might contribute to the existence of a price disparity in
the two markets.
The customary reason for international price disparity between is exchange rate
risk. Historically, there has been very little exchange rate risk because the Hong Kong
Dollar has been closely pegged to the US Dollar. Between 1/2/2003 and 3/28/2008, the
exchange rate ranged from 7.71 to 7.83 HKD per 1 USD. However a possible
explanation behind the price disparity is increased expected exchange rate risk. China is
largely an export country with close ties to the US, so it would make sense for Hong
Kong to continue to keep their currency pegged to the US Dollar. If the value in USD
were to fall without an equal fall in that of HKD, then Chinese companies would be
18
forced to charge higher prices on exports to American consumers. The recent weakness
of the dollar may have caused investors to speculate about a change in the monetary
policy. If investors were expecting that the HKD would rise from its artificially low
position, exchange rate in terms of HKD/USD would have fallen. In the four weeks after
the announcement, the exchange rate did fall from 7.81 to 7.75, or 0.77%. Exchange rate
risk would offer an explanation for any slight price disparity. Another possible reason for
a price disparity is a high monitoring cost or transaction costs. Neither of those hold in
this market however. The monitoring costs are very low since international stock price
quotes are free and it is possible to develop a program that scans for any potential price
inefficiency. Transaction costs will also be low since I assume many international
investors are already positioned in both the HKEX and NYSE markets. So based on this
intuition, we can only expect to see very small price disparities between the two markets.
III. DATA
In the first half of this paper I analyzed the daily pricing changes in the Hong
Kong stock market. The Hang Seng Index, which was started on November 24, 1969, is
a free-float capitalization-weighted index of 40 companies. The stock index currently
represents about 65 percent of the total market capitalization of the Stock Exchange of
Hong Kong. Daily adjusted closing price data of the Hang Seng Index was downloaded
from finance.yahoo.com for the dates 12/31/1986 through 1/25/2008. The first order
differences or return series are computed using the first difference of logarithmic closing
prices:
19
(1)
where pt represents the price at time t. A summary of the price data and first order
differences or price changes of these daily stock market price indices is found in Table
3.1.
TABLE 3.1 Basic Statistics of Index Prices
HIS ΔHSI N 5223 5222
Mean 10078.95 0 Std Dev 5522.264 0.007
In the second half of this paper, I analyze the daily prices of 14 stocks listed on
both the Hong Kong stock market and the New York stock exchange with Level III
ADRs during the time period January 1st, 2003 through March 31st, 2008. Note that
CTEL is an exception because it is actually listed on the NASDAQ rather than the NYSE.
Historical daily stock price quotes were retrieved on April 18th, 2008 from
finance.yahoo.com. The daily HKD (Hong Kong Dollar) to USD (U.S. Dollar) exchange
rate data came from the Federal Reserve Bank of St. Louis and was retrieved on the same
day as all of the stock price quotes. All of the companies studied needed to be listed on
both exchanges throughout the period and needed to not have any significant missing
data. Since the stocks examined are all Level III ADRs, we are guaranteed to have liquid
markets. Three additional companies currently listed on both exchanges did not meet my
criteria because their IPOs occurred after 2003. Another company failed to meet the
criteria because finance.yahoo.com did not have the full data set for the given time frame.
The list of the companies studied is shown in Table 3.2.
)log(log 1 ttt ppr
20
TABLE 3.2 Companies listed on both HKEX and NYSE
H-Share (HKEX) ADR (NYSE) H/ADR Company Code Listing date Code Listing date Ratio Aluminum Corp. of China Ltd. 2600.HK 12/12/2001 ACH 12/11/2001 25 APT Satellite Holdings Ltd. 1045.HK 12/18/1996 ATS 12/17/1996 8 China Eastern Airlines Corp.Ltd. 0670.HK 2/5/1997 CEA 2/4/1997 100 China Mobile Ltd. 0941.HK 10/23/1997 CHL 10/22/1997 5 China Petroleum & Chemical Corp. 0386.HK 10/19/2000 SNP 10/18/2000 100 China Southern Airlines Co. Ltd. 1055.HK 7/31/1997 ZNH 7/30/1997 50 China Telecom Corp. Ltd. 0728.HK 11/15/2002 CHA 11/14/2002 100 China Unicom 0762.HK 6/22/2000 CHU 6/21/2000 10 City Telecom Ltd. 1137.HK 8/4/1997 CTEL 11/3/1999 20 Guangshen Railway Co.Ltd. 0525.HK 5/14/1996 GSH 5/13/1996 50 Huaneng Power International Inc. 0902.HK 1/22/1998 HNP 10/6/1994 40 PetroChina Co. Ltd. 0857.HK 4/7/2000 PTR 4/6/2000 100 Sinopec Shanghai Petrochemical Co. Ltd. 0338.HK 7/26/1993 SHI 7/26/1993 100 Yanzhou Coal Mining Co. Ltd. 1171.HK 4/1/1998 YZC 3/31/1998 50
The daily returns of share prices for the 14 companies were computed using the
first difference of logarithmic closing prices. I chose to examine daily data rather than
weekly or monthly returns in order observe the finer scale dynamics of the time-series
data and present a practical trading rule.
Due to market holidays and a few insignificant holes in the dataset provided by
finance.yahoo.com, there were some issues with the merged time series data set. US-
only and Hong Kong-only holidays caused some of the time series points to have prices
for one exchange and but no prices for the other. To deal with this issue, I chose to insert
new entries for the missing dates, where prices were set to the closing prices of the
previous open day and volume was set to zero. For missing exchange rate data, I
interpolated this value using the two closest exchange rates in the time series. Over the
21
five years of data, there were only seven occurrences in which missing daily exchange
rate data had to be interpolated. Considering that the exchange rates were relatively
stable, the missing data points should not have had a material effect on the validity of the
overall data.
TABLE 3.3 Basic Statistics of Stock Prices
ACH ATS CEA HKEX HKEX$ NYSE HKEX HKEX$ NYSE HKEX HKEX$ NYSE N 1279 1279 1304 1126 1126 1312 1212 1212 1217 Mean 6.344 20.369 20.347 1.756 1.804 1.746 1.957 25.137 25.338 Std Dev 4.855 15.588 15.560 0.456 0.468 0.477 1.662 21.356 21.932 CHA CHL CHU HKEX HKEX$ NYSE HKEX HKEX$ NYSE HKEX HKEX$ NYSE N 1292 1292 1319 1296 1296 1318 1292 1292 1318 Mean 2.981 38.293 38.372 46.197 29.664 29.423 8.031 10.313 10.353 Std Dev 1.229 15.769 15.856 34.523 22.156 22.182 3.645 4.674 4.696 CTEL GSH HNP HKEX HKEX$ NYSE HKEX HKEX$ NYSE HKEX HKEX$ NYSE N 1276 1276 1271 1296 1296 1317 1294 1294 1318 Mean 1.512 3.884 3.807 3.053 19.604 19.672 6.714 34.495 27.660 Std Dev 0.712 1.830 1.865 1.543 9.888 9.902 2.035 10.457 7.492 PTR SHI SNP HKEX HKEX$ NYSE HKEX HKEX$ NYSE HKEX HKEX$ NYSE N 1291 1291 1317 1296 1296 1317 1290 1290 1316 Mean 6.187 79.480 79.919 3.062 39.334 40.003 4.299 55.204 55.711 Std Dev 3.865 49.638 49.461 1.249 16.046 15.807 2.676 34.333 34.384 YZC ZNH HKEX HKEX$ NYSE HKEX HKEX$ NYSE N 1293 1293 1299 1289 1289 1309 Mean 7.677 49.295 41.198 3.540 22.727 22.818 Std Dev 3.388 21.726 23.099 2.245 14.418 14.611
In order to compare the prices of Hong Kong stocks to US stocks, the Hong Kong
stock prices were converted to USD using the daily exchange rate, and then divided by
the corresponding HK shares/US shares factor found in Table 3.2. The H-share-to-ADR
22
ratio is determined when the company first issues ADRs and represents the value of one
share of the respective company traded on the HKEX in relation of one share traded on
the NYSE. Table 3.3 displays a summary of the characteristics of the entire data set for
HKEX, USD-adjusted HKEX, and NYSE prices, respectively, of all 14 stocks. The
descriptive statistics show that the mean and standard deviation of stock prices in Hong
Kong and the US were very similar. For most of the stocks, the standard deviation of the
stock listed on the NYSE was slightly higher than the corresponding listing on the
HKEX. This suggests that the US market is actually somewhat riskier than the HK
market for those dual-listed stocks. All of the unexpected findings are attributable to
problems with incorrect data from finance.yahoo.com. For example, the Hong Kong
stock price data was incorrect for HNP from 2003 to 2004 and for YZC from 2003 to
2005. It is most likely that these problems occurred because the stock prices were not
properly adjusted for stock splits. In addition, data is missing for CEA between May
22nd, 2007 and August 31st, 2007. This missing data will exclude CEA from the tests on
the 8-week period surrounding the announcement in August of 2007.
IV. METHODOLOGY
The aim of this paper is to examine how the two aforementioned events affected
the efficiency of the Hong Kong stock market. The Geary run test is used as a way to
find serial dependence before and after the opening of the options market. The Granger
Causality test can determine the predictive power of price of one stock on the price of
another.
23
4.1. Geary Run Test. The Geary run test computes the number of sequences of positive
and negative returns and compares it against its sampling distribution under the random
walk hypothesis (Campbell et al., 1997). The total expected number of runs under
successive outcomes are independent if distributed normally with the following mean and
standard deviation:
(5)
(6)
where n is the number of runs of type i. The test for serial dependence is carried out by
comparing the observed number of runs with the expected number of runs under the
random walk hypothesis. If the observed number of runs is much less than the expected
number, this would mean that the market exhibits mean reversion; a rise in the stock
market one day would generally be followed by a drop the next day.
4.2. Granger Causality Test. The test for Granger causation must be performed on
stationary data. In order to describe these tests in detail, it is necessary to define a few
terms first. A stationary process is a random process where all of its statistical properties
do not vary with time. The mean and the variance of the underlying process are constant
and the autocovariances depend only on the time lag. A process whose statistical
properties do change with time is called nonstationary. A series which is stationary after
being differentiated once is said to be integrated of order 1 and is denoted by I(1). A
linear stochastic process has a unit root if 1 is a root of the process’s characteristic
equation.
NnNN e
i i
12)1(
21
2
3
1
3
13323
1
)1()(2)]1([
NNNnNNNn
i i iii
24
The Dickey-Fuller test determines whether or not a unit root is present in an
autoregressive model (Dickey and Fuller, 1979). Testing for a unit root is based on the
following first-order autoregressive, AR(1), model:
(2)
where is the white-noise error term and the initial value, , is constant. If , a
unit root is present and thus the model would be non-stationary. If then the series
is stationary. By subtracting from both sides of the equation ,
We have:
(3)
(4)
We are testing whether or not the stock price follows a pure random walk. Our null
hypothesis is that 0 in which case the time series has a unit root. The alternate
hypothesis 0 would lead to a conclusion that the time series is stationary. The
weakness of this test is that it will detect only linear forms of dependencies.
Granger testing is a common method of investigating causal relationships
(Granger, 1969). A variable X Granger-causes Y if Y can be better predicted using the
histories of both X and Y than just using just the history of Y. Stated differently: if X-
values provide statistically significant information about the future values of Y, then X
Granger-causes Y. The Granger Causality Test is an F-test of the joint significance of the
other variable in a regression that includes a lags of the dependent variable. Note that
Granger causality does not imply true causality. A problem with this test is that it may
11 tt yay
0y 11 a
11 a
1ty
111 )1( ttt yayy
11 tt yay
1tt yy
25
produce misleading results when the true relationship involves three or more variables. If
X and Y are both caused by a third factor, they may have no true relationship with each
other, yet give positive results in a Granger test.
4.4. Arbitrage Trading Rule. The trading strategy between HKEX and NYSE is as
follows: if the price differential between the two exchanges is greater than x%, first place
an order in the open exchange. If the open exchange has a price premium, then the order
would be a short sale; if the exchange has a price discount, then the order would be to
buy shares. Once the other stock market opens, we should note if the differential still
surpasses the threshold level. If so, then complete the corresponding arbitrage position
and wait for the differential to converge to zero before exiting both positions. Otherwise,
exit the original position the next available trading day. Because the two stock exchanges
are not open at the same time, this strategy will not be risk-free. Since the return times
series is stationary, the risk of holding a security for a day is expected to be very small.
Unlike other global arbitrage trading rules, however, exchange rate risk is not much of a
factor here since the Hong Kong dollar is pegged to the US dollar. I am testing to see if
this strategy could have achieved a period of above average returns immediately after
China’s announcement.
V. EMPIRICAL RESULTS AND ANALYSIS
5.1. Stock Options Market. As discussed in the previous section, the Geary runs test
can determine whether or not a there is serial dependence in a time series, depending on
how closely the observed number of positive and negative runs compares to the expected
26
TABLE 5.1 Geary Run tests for daily data
Period Observed # of Runs Expected # of Runs Negative Positive 1987-2007 2567 4628 1278 1212
1988-1995 940 1773 474 439 (Pre-market) 1996-2003 968 1734 490 484
(Post-market)
number of runs distributed under the random walk hypothesis. The test results are shown
in Table 5.1. On the first examination of these results, it is clear that the observed
number of runs is less than the number of runs expected if the stock followed under a
random walk hypothesis. The pairs of positive and negative runs are all very close
together. From this test we can conclude that the prices of the Hang Seng Index do not
follow a random walk. Comparing the results before and after the establishment of the
options market, I found that the observed number of runs is closer to the expected
number, but not enough to be significant.
5.2. Announcement Effect. I ran two Dickey-Fuller tests for unit roots on each of the
14 companies: one using the NYSE price data and another using the exchange-adjusted
HKEX price data of all 14 companies. For most of the stocks, there was not enough
evidence to reject the null hypothesis; the prices were found to be nonstationary. ATS
was the only stock ticker that had stationary prices for both the US and Hong Kong
listings. Table 5.2 displays the test results for all 28 unit root tests of price series.
Almost all of the stocks were found to be non-stationary. Since the Granger test can be
run only on a stationary time series, the Dickey-Fuller test was run on the return price
data, or first difference of stock prices. The results in Table 5.3 show that the return
27
series for the every stock was found to be stationary. In conjunction with the stationary
return series, the daily closing price graphs in the Appendix provide enough evidence to
assert that Hong Kong stock prices and US prices are cointegrated.
TABLE 5.2 Unit Root of Stock Prices
US HK Code N Test Statistic p-value Stationary Code N Test Statistic p-value Stationary ACH 1053 -0.814 0.8150 No 2600.HK 1028 -0.974 0.7627 No ATS 1057 -2.901 0.0453 Yes 1045.HK 908 -2.878 0.0480 Yes CEA 980 0.267 0.9758 No 0670.HK 976 0.328 0.9786 No CHA 1064 -1.508 0.5297 No 0728.HK 1038 -0.510 0.8900 No CHL 1065 0.647 0.9887 No 0941.HK 1042 1.830 0.9984 No CHU 1063 0.126 0.9678 No 0762.HK 1038 1.421 0.9972 No CTEL 1023 -2.491 0.1177 No 1137.HK 1028 -0.771 0.8275 No GSH 1065 -2.418 0.1367 No 0525.HK 1042 -2.295 0.1735 No HNP 1064 -2.254 0.1874 No 0902.HK 1038 -3.135 0.0241 Yes PTR 1064 -0.853 0.8031 No 0857.HK 1039 -0.536 0.8848 No SHI 1064 -1.500 0.5334 No 0338.HK 1042 -1.554 0.5065 No SNP 1062 -0.159 0.9433 No 0386.HK 1037 0.242 0.9745 No YZC 1049 -1.598 0.4844 No 1171.HK 1040 -1.858 0.3519 No ZNH 1056 0.113 0.9670 No 1055.HK 1036 0.293 0.9770 No
TABLE 5.3 Unit Root of First Difference of Stock Prices
US HK Code N Test Statistic p-value Stationary Code N Test Statistic p-value Stationary ΔACH 1052 -34.148 0.0000 Yes Δ2600.HK 1027 -31.854 0.0000 Yes ΔATS 1056 -35.642 0.0000 Yes Δ1045.HK 907 -33.919 0.0000 Yes ΔCEA 979 -31.430 0.0000 Yes Δ0670.HK 975 -35.312 0.0000 Yes ΔCHA 1063 -35.889 0.0000 Yes Δ0728.HK 1037 -32.663 0.0000 Yes ΔCHL 1064 -37.990 0.0000 Yes Δ0941.HK 1041 -33.136 0.0000 Yes ΔCHU 1062 -34.313 0.0000 Yes Δ0762.HK 1037 -31.620 0.0000 Yes ΔCTEL 1022 -36.553 0.0000 Yes Δ1137.HK 1027 -29.841 0.0000 Yes ΔGSH 1064 -37.558 0.0000 Yes Δ0525.HK 1041 -35.963 0.0000 Yes ΔHNP 1063 -36.254 0.0000 Yes Δ0902.HK 1037 -33.348 0.0000 Yes ΔPTR 1063 -36.861 0.0000 Yes Δ0857.HK 1038 -32.602 0.0000 Yes ΔSHI 1061 -36.070 0.0000 Yes Δ0338.HK 1036 -33.408 0.0000 Yes ΔSNP 1061 -36.476 0.0000 Yes Δ0386.HK 1036 -34.114 0.0000 Yes ΔYZC 1048 -32.610 0.0000 Yes Δ1171.HK 1039 -32.873 0.0000 Yes ΔZNH 1055 -33.119 0.0000 Yes Δ1055.HK 1035 -32.324 0.0000 Yes
28
I then ran the Granger Causality test to determine whether or not one stock
exchange’s price time series is useful in forecasting the corresponding stock in the other
market. The Granger causality test was run using a lag length of four, based on the
bivariate time series model:
(5)
(6)
where and . represent the return series for one of the two stocks at time t.
Because daily returns were used, the choice of lag length was crucial. A weakness of the
Granger Causality test is that the chosen lag length could have a significant impact on
results. After running the tests with four lags, I found that one lag was enough. This was
an expected result because of the nature of the two markets: prices in the US market are
correlated to both the previous day price and the previous HKEX price. From the results
in Table 5.4, it is interesting to note that the Hong Kong-listed SNP was the only stock
that did not hold significant forecast power over their respective counterparts. Thus, we
can conclude that price discovery exists between the HKEX and NYSE. Because both
markets contribute to the price discovery process, price changes in either market have a
significant impact on prices in the other.
Finally, the last test I ran was the simulation of an arbitrage trading strategy using
the historical data of the fourteen different companies. I accounted for closed market
days by executing only when both Hong Kong and US markets had volume above zero.
Due to size of these companies and high volume of trades on average, liquidity concerns
were not expected to be a problem. This test was unable to conclusively find any
44332211 ttttt XXXXX
144332211 tttt YYYY
44332211 ttttt YYYYY
244332211 tttt XXXX
tX tY
29
arbitrage opportunities. After a quick glance at the plots of the 14 stocks in the
Appendix, it becomes evident that there were not any significant profit opportunities
across the studied stocks immediately after the announcement. If a small transaction fee
per trade were taken into account, the arbitrage strategy would have lost money during
the eight-week period.
TABLE 5.4 Granger Causality Test for ADRs
Equation Excluded χ2 df Prob > χ2 Causality ACH 2600.HK 68.549 4 0.000 Yes ATS 1045.HK 16.335 4 0.003 Yes CEA 0670.HK 55.867 4 0.000 Yes CHA 0728.HK 20.551 4 0.000 Yes CHL 0941.HK 30.708 4 0.000 Yes CHU 0762.HK 13.317 4 0.010 Yes CTEL 1137.HK 39.054 4 0.000 Yes GSH 0525.HK 8.774 4 0.067 Yes HNP 0902.HK 16.512 4 0.002 Yes PTR 0857.HK 17.626 4 0.001 Yes SHI 0338.HK 8.281 4 0.082 Yes SNP 0386.HK 5.327 4 0.255 No YZC 1171.HK 8.382 4 0.079 Yes ZNH 1055.HK 43.777 4 0.000 Yes
TABLE 5.5 Granger Causality Test for H-Shares
Equation Excluded χ2 df Prob > χ2 Causality 2600.HK ACH 44.807 4 0.000 Yes 1045.HK ATS 25.629 4 0.000 Yes 0670.HK CEA 219.39 4 0.000 Yes 0728.HK CHA 84.191 4 0.000 Yes 0941.HK CHL 187.03 4 0.000 Yes 0762.HK CHU 117.03 4 0.000 Yes 1137.HK CTEL 9.2146 4 0.056 Yes 0525.HK GSH 84.492 4 0.000 Yes 0902.HK HNP 69.014 4 0.000 Yes 0857.HK PTR 316.49 4 0.000 Yes 0338.HK SHI 149.13 4 0.000 Yes 0386.HK SNP 411.61 4 0.000 Yes 1171.HK YZC 123.4 4 0.000 Yes 1055.HK ZNH 174.06 4 0.000 Yes
30
The one remaining question is determining the reason behind a high price increase
after the announcement. The law of one price states the following: “In an efficient
market all identical goods must have only one price.” In the Hong Kong and US markets,
the identical goods were shares of dual-listed Chinese companies and the prices were
found to be the same. As a result of the straightforward opportunity to undertake
arbitrage between the two markets, the price series for the same goods listed on both
exchanges never exhibited much price disparity even in a volatile market. There were no
changes to the intrinsic values of the companies after the China made the announcement
that more capital would be entering the Hong Kong stock exchange. Therefore, the
reason behind a huge increase in stock price across all of the Hong Kong stocks must
have been supply and demand. In the stock market, the short-term supply of stocks is
limited. A big increase in expected demand from optimistic investors is the only logical
explanation for this story.
VI. CONCLUSION
This thesis investigated price discovery and efficiency in the Hong Kong stock
exchange after a large event or announcement. The paper considered large events
because it attempted to find situations in which the Adaptive Markets Hypothesis would
provide a better explanation for inefficiencies than the Efficient Markets Hypothesis.
Technological innovations and high price fluctuations appear to be situations where one
would likely find temporary profitable opportunities. Unfortunately, the two events
31
studied in the Hong Kong market did not display the level of inefficiency expected in
order to adopt AMH.
The introduction of the stock exchange was found to slightly increase the
efficiency of the Hong Kong stock exchange, but empirical analysis was unable to
conclude this with any certainty. An examination of the time period following China’s
announcement found a high level of uncertainty and price movement in both the US-
listed and Hong Kong-listed group of stocks. Results from the Granger Causality test and
the arbitrage test supports the theory that there is a high level of price discovery between
the two markets in spite of the volatile price movements.
In my attempt to determine why the stock prices increased so drastically after
China’s announcement, I concluded that the expected participation of optimistic investors
in the stock market would create a shift in the demand curve which temporarily results in
a stock price increase. Further research is needed to examine whether or not the proposed
supply and demand story really holds. In the long run, companies are able to sell off
shares that they store in treasury to take advantage of the higher stock price. We would
expect to see stock prices sinking back down to their pre-announcement levels.
32
REFERENCES
Banz, R. "The relationship between return and market value of common stocks." Journal of Financial Economics, 9 (1981): 3-18.
Barberis, N. and R. H. Thaler. "A Survey of Behavioral Finance." September (2002). Available at SSRN: http://ssrn.com/abstract=327880. DOI: 10.2139/ssrn.327880.
Basu, S. "Investment performance of common stocks in relation to their price-earning ratios: A test of the efficient market hypothesis." Journal of Finance, 32 (1977): 663-682.
Bhardwaj, R. K. and L. D. Brooks. "The January anomaly: Effects of low share price, transaction costs, and bid-ask bias." Journal of Finance, 47 (1992): 553-575.
Brock, W., J. Lakonishok and B. LeBaron. "Simple Technical Trading Rules and the Stochastic Properties of Stock Returns." Journal of Finance, 47 (1992): 1731-1764.
Campbell, J. Y. and R. J. Shiller. "Stock prices, earnings and expected dividends." Journal of Finance, 43 (1988): 661-676.
Campbell, J.Y., Lo A.W. and MacKinlay A.C. (1997) The Econometrics of Financial Markets. Princeton University Press, New Jersey.
Chan, K.L. and K.H. Kwok, 2005, “Market Segmentation and Share Price Premium: Evidence from Chinese Stock Markets,” Journal of Emerging Market Finance 4:1.
Chang, Eric C. and Yu, Yinghui, "Short-Sales Constraints and Price Discovery: Evidence from the Hong Kong Market" (March 15, 2004). EFA 2004 Maastricht Meetings Paper No. 3641. Available at SSRN: http://ssrn.com/abstract=565330
Chen, K. (2003), The Price Relationships among the Chinese Company’s Shares Listing Domestically and Abroad, Göteborg, Graduate Business School
Daniel, K., D. Hirshleifer and A. Subrahmanyam. "Covariance Risk, Mispricing, and the Cross Section of Security Returns." Journal of Finance, ??? (2000): ???-???
Dickey, D. A. and W. A. Fuller. "Distribution of the Estimators for Autoregressive Time Series with a Unit Root." Journal of the American Statistical Association, 74 (1979): 427-431.
Dow, D. and G. Gorton. Stock Market Efficiency and Economic Efficiency: Is There a Connection? Journal of Finance. American Finance Association. 52 (3), July (1997): 1087-1129.
33
Eleswarapu, V. R. and M. R. Reinganum. "The seasonal behavior of the liquidity premium in asset pricing." Journal of Financial Economics, 34 (1993): 373-386.
Engle, R.F. and C.W. Granger, 1987, “Co-integration and Error Correction: Representation, Estimation, and Testing,” Econometrica, vol.55, pp. 251-76.
Fama, E. F. "Efficient Capital Markets: A Review of Theory and Empirical Work." The Journal of Finance, (1970): 383-417.
Fama, E. F. "Market Efficiency, Long-Term Returns, and Behavioral Finance." February (1997). Available at SSRN: http://ssrn.com/abstract=15108. DOI: 10.2139/ssrn.15108.
French, K. "Stock returns and the weekend effect." Journal of Financial Economics, 8 (1980): 55-69.
Grossman, S. J. a. J. E. S. "On the Impossibility of Informationally Efficient Markets." American Economic Review, 70(3),June (1980): 393-408.
Gultekin, M. and B. Gultekin. "Stock market seasonality: International evidence." Journal of Financial Economics, 12 (1983): 469-481.
Guo, L., and Tang, L., 2006, “Cost of Capital and Liquidity of Cross-Listed Chinese Companies,” Presented at Financial Management Association, Salt Lake City, Utah.
Harris, L. and E. Gurel. "Price and volume effects associated with changes in the S&P 500 list: New evidence for the existence of price pressures." Journal of Finance, 41 (1986): 815-829.
Hirshleifer, D. a. T. S. "Good day sunshine: Stock returns and the weather." SSRN Working Paper, Forthcoming, Journal of Finance, (2001).
Hong, H. a. J. C. S. "A Unified Theory of Underreaction, Momentum Trading, and Overreaction in Asset Markets." Journal of Finance, 54(6), (1999).
Kamara, A. "New evidence on the Monday seasonal in stock returns." Journal of Business, 70 (1997): 63-84.
Lakonishok, J., A. Shleifer and R. Vishny. "Contrarian Investment, Extrapolation, and Risk." Journal of Finance, 49 (1994): 1541–78.
Lo, A. W. "The Adaptive Markets Hypothesis: Market Efficiency from an Evolutionary Perspective." Journal of Portfolio Management, (2004): Forthcoming Available at SSRN: http://ssrn.com/abstract=602222.
34
Lo, A. W. and A. C. MacKinlay. "Stock Market Prices Do Not Follow Random Walks: Evidence from a Simple Specification Test." The Review of Financial Studies, 1(1), (1988): 41-66.
Los, C. A. "Nonparametric Efficiency Testing of Asian Stock Markets Using Weekly Data." (1998). Centre for Research in Financial Services Working Paper No. 99-01. Available at SSRN: http://ssrn.com/abstract=143950. DOI: 10.2139/ssrn.143950.
Odean, T. "Volume, volatility, price and profit when all traders are above average." Journal of Finance, 53 (1998): 1887-1934.
Ratner, M. and R. Leal. "Tests of Technical Trading Strategies in the Emerging Equity Markets of Latin America and Asia." Journal of Banking and Finance, 23 (1999): 1887-1905.
Reinganum, R. M. "Misspecification of the capital asset pricing: Empirical anomalies based on earnings yields and market values." Journal of Financial Economics, 12 (1981): 89-104.
Rozeff, M. S. and W. R. Kinney. "Capital market seasonality: The case of stock returns." Journal of Financial Economics, 3 (1976): 379-402.
Samuelson, P. "Proof that properly anticipated prices fluctuate randomly." Industrial Management Review, 6 (1965): 41-49.
Saunders, E. M. "Stock prices and Wall Street weather." American Economic Review, 83 (1993): 1337-1345.
Seyhun, H. N. "Insiders’ profits, costs of trading, and market efficiency." Journal of Financial Economics, 16 (1986): 189-212.
Seyhun, H. N. Investment Intelligence from Insider Trading. (1998) Cambridge, MA, MIT Press.
Shleifer, A. "Do demand curves for stocks slope down." Journal of Finance, 41 (1986): 579-590.
Shleifer, A. and R. Vishny. "The limits of arbitrage." Journal of Finance, 52 (1990): 35-55.
Sullivan, R., A. Timmerman and H. White. "Data-Snooping, Technical Trading Rule performance, and the Bootstrap." Journal of Finance, 54 (1994): 1647-1691.
Tobin, J. "On the Efficiency of the Financial System." Lloyd’s Banking Review, July (1982).
35
Wang, S.Y. and J. Li, “Location of Trade, Ownership Restrictions, and Market Illiquidity: Examining Chinese A- and H-Shares.” Available at SSRN: http://ssrn.com/abstract=410702 or DOI: 10.2139/ssrn.410702
36
37
38
39
40
41
42